315 lines
11 KiB
C
315 lines
11 KiB
C
/* md_demo.c - Markdown 渲染效果演示
|
|
*
|
|
* 一个独立的小程序: 把内置示例 (或你指定的文件 / stdin)喂进 ligent 的
|
|
* 流式 Markdown 渲染器, 直观看到每种语法的渲染结果.
|
|
*
|
|
* 编译: make md_demo
|
|
* 运行: ./build/md_demo
|
|
*/
|
|
#include "markdown.h"
|
|
#include "util.h"
|
|
|
|
#include <stdio.h>
|
|
#include <stdlib.h>
|
|
#include <string.h>
|
|
#include <time.h>
|
|
#include <unistd.h>
|
|
|
|
/* ------------------------------------------------------------------ */
|
|
/* 内置示例: 把支持的语法和边界情况都覆盖一遍 */
|
|
/* ------------------------------------------------------------------ */
|
|
|
|
static const char *SAMPLE =
|
|
"# ligent 渲染演示\n"
|
|
"\n"
|
|
"## 行内样式\n"
|
|
"\n"
|
|
"这是 **粗体**、*斜体*、_下划线斜体_ 和 `行内代码` 的混排效果. \n"
|
|
"标识符不会被误伤: foo_bar_baz、__init__、arr[i]、a_b_c 都保持原样. \n"
|
|
"链接只高亮方括号: [官方文档](https://example.com/doc) 后面还有普通文字. \n"
|
|
"图片保留原文: \n"
|
|
"\n"
|
|
"## 列表与引用\n"
|
|
"\n"
|
|
"- 减号列表项\n"
|
|
"- 第二项\n"
|
|
"* 星号也能当列表\n"
|
|
"+ 加号也可以\n"
|
|
"\n"
|
|
"> 这是一段引用, 前缀是暗色的竖线. \n"
|
|
"> 引用可以有多行. \n"
|
|
"\n"
|
|
"---\n"
|
|
"\n"
|
|
"## 代码块 (整块不渲染)\n"
|
|
"\n"
|
|
"```c\n"
|
|
"#include <stdio.h>\n"
|
|
"int main(void) {\n"
|
|
" // **这里不会加粗**, [方括号] 也不会高亮\n"
|
|
" printf(\"hello %s\\n\", \"world\");\n"
|
|
" return 0;\n"
|
|
"}\n"
|
|
"```\n"
|
|
"\n"
|
|
"行内代码同理: `a ** b` 里的星号不会被当成粗体标记. \n"
|
|
"\n"
|
|
"## 表格\n"
|
|
"\n"
|
|
"| 语言 | 类型系统 | 典型用途 | 份额 |\n"
|
|
"|:-----|:--------:|----------|-----:|\n"
|
|
"| C | 静态弱类型 | 系统编程 | 11.4% |\n"
|
|
"| Rust | 静态强类型 | 系统编程 | 3.2% |\n"
|
|
"| Python | 动态强类型 | 脚本 / AI | 23.1% |\n"
|
|
"\n"
|
|
"| 单列表格 |\n"
|
|
"|---|\n"
|
|
"| 也能渲染 |\n"
|
|
"\n"
|
|
"## 边界情况\n"
|
|
"\n"
|
|
"乘法星号原样保留: 2 * 3 * 4 = 24\n"
|
|
"管道命令不会被误判成表格: ls -la | grep foo\n"
|
|
"未闭合的 **粗体标记 会在行尾被复位\n"
|
|
"#井号后没有空格就不是标题\n"
|
|
"####### 七个井号也不是标题\n"
|
|
"\n";
|
|
|
|
/* ------------------------------------------------------------------ */
|
|
/* 渲染到内存 (用于 --compare 自检) */
|
|
/* ------------------------------------------------------------------ */
|
|
|
|
static char *render_to_mem(const char *src, int enabled, size_t chunk) {
|
|
char *buf = NULL;
|
|
size_t size = 0;
|
|
FILE *f = open_memstream(&buf, &size);
|
|
if (!f)
|
|
return NULL;
|
|
|
|
MdRenderer *r = md_new(enabled);
|
|
md_set_out(r, f);
|
|
|
|
if (chunk == 0) {
|
|
md_feed(r, src);
|
|
} else {
|
|
size_t len = strlen(src);
|
|
for (size_t i = 0; i < len; i += chunk) {
|
|
size_t n = len - i < chunk ? len - i : chunk;
|
|
char *tmp = xstrndup(src + i, n);
|
|
md_feed(r, tmp);
|
|
free(tmp);
|
|
}
|
|
}
|
|
md_finish(r);
|
|
md_free(r);
|
|
fclose(f);
|
|
return buf;
|
|
}
|
|
|
|
/* ------------------------------------------------------------------ */
|
|
/* 渲染到终端 (可带分片与延时, 模拟 SSE 流式) */
|
|
/* ------------------------------------------------------------------ */
|
|
|
|
static void render_to_tty(const char *src, int enabled, size_t chunk,
|
|
long delay_ms) {
|
|
MdRenderer *r = md_new(enabled);
|
|
md_set_out(r, stdout);
|
|
|
|
size_t len = strlen(src);
|
|
if (chunk == 0)
|
|
chunk = len ? len : 1;
|
|
|
|
struct timespec ts;
|
|
ts.tv_sec = delay_ms / 1000;
|
|
ts.tv_nsec = (delay_ms % 1000) * 1000000L;
|
|
|
|
for (size_t i = 0; i < len; i += chunk) {
|
|
size_t n = len - i < chunk ? len - i : chunk;
|
|
char *tmp = xstrndup(src + i, n);
|
|
md_feed(r, tmp);
|
|
free(tmp);
|
|
fflush(stdout);
|
|
if (delay_ms > 0)
|
|
nanosleep(&ts, NULL);
|
|
}
|
|
md_finish(r);
|
|
md_free(r);
|
|
}
|
|
|
|
/* ------------------------------------------------------------------ */
|
|
/* 参数 */
|
|
/* ------------------------------------------------------------------ */
|
|
|
|
static void usage(FILE *out) {
|
|
fputs("md_demo - 演示 ligent 的流式 Markdown 渲染效果\n"
|
|
"\n"
|
|
"用法: md_demo [选项] [文件...]\n"
|
|
"\n"
|
|
"不指定文件时使用内置示例; 文件名为 - 或 stdin "
|
|
"不是终端时从标准输入读取. \n"
|
|
"\n"
|
|
"选项: \n"
|
|
" -p, --plain 关闭渲染, 输出原始 Markdown (对照用)\n"
|
|
" --color 强制开启渲染 (默认仅在输出到终端时开启)\n"
|
|
" -c, --chunk N 按每 N 字节一片喂入 (默认 0 = 一次性喂完)\n"
|
|
" -d, --delay MS 每片之间延时毫秒 (默认 0)\n"
|
|
" -s, --stream 流式预设: 等价于 -c 5 -d 12\n"
|
|
" --compare 自检: 把同一段文本按多种分片方式渲染, "
|
|
"比对结果是否一致\n"
|
|
" -h, --help 显示本帮助\n"
|
|
"\n"
|
|
"例子: \n"
|
|
" md_demo 看渲染效果\n"
|
|
" md_demo --plain 看原始 Markdown, 对照差异\n"
|
|
" md_demo -c 5 -d 12 模拟 SSE 流式 (标记会被随机切断)\n"
|
|
" md_demo --compare 验证分片不变性\n"
|
|
" md_demo README.md 渲染任意 Markdown 文件\n"
|
|
" echo '# hi **there**' | md_demo\n",
|
|
out);
|
|
}
|
|
|
|
static char *read_stream(FILE *f) {
|
|
StrBuf sb;
|
|
sb_init(&sb);
|
|
char buf[8192];
|
|
size_t n;
|
|
while ((n = fread(buf, 1, sizeof buf, f)) > 0)
|
|
sb_append(&sb, buf, n);
|
|
return sb_detach(&sb);
|
|
}
|
|
|
|
int main(int argc, char **argv) {
|
|
int plain = 0, force_color = 0, compare = 0;
|
|
long delay_ms = 0;
|
|
size_t chunk = 0;
|
|
const char *file = NULL;
|
|
|
|
for (int i = 1; i < argc; i++) {
|
|
const char *a = argv[i];
|
|
if (strcmp(a, "-h") == 0 || strcmp(a, "--help") == 0) {
|
|
usage(stdout);
|
|
return 0;
|
|
} else if (strcmp(a, "-p") == 0 || strcmp(a, "--plain") == 0) {
|
|
plain = 1;
|
|
} else if (strcmp(a, "--color") == 0) {
|
|
force_color = 1;
|
|
} else if (strcmp(a, "-c") == 0 || strcmp(a, "--chunk") == 0) {
|
|
if (++i >= argc) {
|
|
fputs("缺少 -c 的参数\n", stderr);
|
|
return 2;
|
|
}
|
|
chunk = (size_t)strtoul(argv[i], NULL, 10);
|
|
} else if (strcmp(a, "-d") == 0 || strcmp(a, "--delay") == 0) {
|
|
if (++i >= argc) {
|
|
fputs("缺少 -d 的参数\n", stderr);
|
|
return 2;
|
|
}
|
|
delay_ms = strtol(argv[i], NULL, 10);
|
|
} else if (strcmp(a, "-s") == 0 || strcmp(a, "--stream") == 0) {
|
|
chunk = 5;
|
|
delay_ms = 12;
|
|
} else if (strcmp(a, "--compare") == 0) {
|
|
compare = 1;
|
|
} else if (a[0] == '-' && a[1] != '\0' && strcmp(a, "-") != 0) {
|
|
fprintf(stderr, "未知选项 %s (--help 查看用法)\n", a);
|
|
return 2;
|
|
} else {
|
|
file = a;
|
|
}
|
|
}
|
|
|
|
/* 取输入 */
|
|
char *src = NULL;
|
|
const char *source_name = "内置示例";
|
|
if (file && strcmp(file, "-") != 0) {
|
|
size_t len = 0;
|
|
src = read_entire_file(file, &len);
|
|
if (!src) {
|
|
fprintf(stderr, "无法读取文件 %s\n", file);
|
|
return 1;
|
|
}
|
|
source_name = file;
|
|
} else if (file || !isatty(fileno(stdin))) {
|
|
src = read_stream(stdin);
|
|
source_name = "标准输入";
|
|
if (src[0] == '\0') {
|
|
/* stdin 是空管道 (例如被 CI / make 调用)时退回内置示例 */
|
|
free(src);
|
|
src = xstrdup(SAMPLE);
|
|
source_name = "内置示例 (stdin 为空)";
|
|
}
|
|
} else {
|
|
src = xstrdup(SAMPLE);
|
|
}
|
|
|
|
int enabled =
|
|
plain ? 0 : (force_color ? 1 : (isatty(fileno(stdout)) ? 1 : 0));
|
|
|
|
if (compare) {
|
|
/* 分片不变性自检: 不同分片大小必须得到完全相同的结果 */
|
|
puts("分片不变性自检 (同一段文本, 不同分片方式, 输出必须逐字节一致)");
|
|
printf("输入: %s (%zu 字节)\n\n", source_name, strlen(src));
|
|
|
|
char *whole = render_to_mem(src, 1, 0);
|
|
size_t sizes[] = {1, 2, 3, 5, 7, 13, 64};
|
|
int all_ok = 1;
|
|
for (size_t k = 0; k < sizeof sizes / sizeof sizes[0]; k++) {
|
|
char *part = render_to_mem(src, 1, sizes[k]);
|
|
int ok = part && whole && strcmp(whole, part) == 0;
|
|
printf(" %-5zu 字节/片 %s\n", sizes[k],
|
|
ok ? "一致 ✓" : "不一致 ✗");
|
|
if (!ok)
|
|
all_ok = 0;
|
|
free(part);
|
|
}
|
|
/* 随机分片 (模拟真实 SSE 的抖动) */
|
|
srand(12345);
|
|
char *buf = NULL;
|
|
size_t bsize = 0;
|
|
FILE *f = open_memstream(&buf, &bsize);
|
|
MdRenderer *r = md_new(1);
|
|
md_set_out(r, f);
|
|
size_t len = strlen(src), i = 0;
|
|
while (i < len) {
|
|
size_t n = (size_t)(rand() % 11) + 1;
|
|
if (i + n > len)
|
|
n = len - i;
|
|
char *tmp = xstrndup(src + i, n);
|
|
md_feed(r, tmp);
|
|
free(tmp);
|
|
i += n;
|
|
}
|
|
md_finish(r);
|
|
md_free(r);
|
|
fclose(f);
|
|
int ok = buf && whole && strcmp(whole, buf) == 0;
|
|
printf(" %-5s %s\n", "随机", ok ? "一致 ✓" : "不一致 ✗");
|
|
if (!ok)
|
|
all_ok = 0;
|
|
free(buf);
|
|
free(whole);
|
|
|
|
printf("\n结论: %s\n", all_ok ? "全部一致, 标记跨分片处理正确. "
|
|
: "存在不一致, 渲染器有 bug. ");
|
|
free(src);
|
|
return all_ok ? 0 : 1;
|
|
}
|
|
|
|
if (isatty(fileno(stdout))) {
|
|
fprintf(stderr, "── 输入: %s 渲染: %s%s ──\n\n", source_name,
|
|
enabled ? "开" : "关", chunk ? " (分片模式)" : "");
|
|
}
|
|
|
|
render_to_tty(src, enabled, chunk, delay_ms);
|
|
|
|
if (isatty(fileno(stdout))) {
|
|
fprintf(stderr, "\n── 演示结束 ──\n");
|
|
if (enabled && chunk == 0)
|
|
fprintf(stderr, "试试 `md_demo -s` 看流式分片下标记被切断时的表现, "
|
|
"或 `md_demo -p` 对照原始 Markdown. \n");
|
|
}
|
|
|
|
free(src);
|
|
return 0;
|
|
}
|